Proc. Natl. Acad. Sci. USAVol. 90, pp. 7275-7279, August 1993Cell Biology

Distribution of the aquaporin CHIP in secretory and resorptiveepithelia and capillary endothelia

(water channel/chorold plexus/ciliary epithelium/bile ducts/intestinal lacteals)

S0REN NIELSEN*, BARBARA L. SMITHt, ERIK ILS0 CHRISTENSEN*, AND PETER AGREtT*Department of Cell Biology, University of Aarhus, DK 8000, Aarhus C, Denmark; and tDepartments of Medicine and Cell Biology/Anatomy, Johns HopkinsUniversity, Baltimore, MD 21205

Communicated by Thomas D. Pollard, May 6, 1993

ABSTRACT The existence of water-selective channels hasbeen postulated to explain the high water permeability oferythrocytes and certain epithelial cells. The aquaporin CHIP(channel-forming integral membrane protein of 28 kDa), amolecular water channel, is abundant in erythrocytes andwater-permeable segments of the nephron. To determinewhether CHIP may mediate transmembrane water movementin other water-permeable epithelia, membranes of multipleorgans were studied by immunoblotting, immunohistochemis-try, and immunoelectron microscopy using afmity-purifiedanti-CHIP IgG. The apical membrane of the choroid plexusepithelium was densely stained, implying a role for CHIP in thesecretion ofcerebrospinal fluid. In the eye, CHIP was abundantin apical and basolateral domains of ciliary epithelium, the siteof aqueous humor secretion, and also in lens epitheilum andcorneal endothelium. CHIP was detected in membranes ofhepatic bile ducts and water-resorptive epithelium of gallbladder, suggesting a role in bile secretion and concentration.CHIP was not detected in glandular epithelium of mammary,salivary, or lacrimal glands, suggesting the existence of otherwater-channel isoforms. CHIP was also not detected within theepithelium of the gastrointestinal mucosa. CHIP was abundantin membranes of intestinal lacteals and continuous capillariesin diverse tissues, including cardiac and skeletal muscle, thusproviding a molecular explanation for the known water per-meability of certain lymphatics and capillary beds. Thesestudies underscore the hypothesis that CHIP plays a major rolein transcellular water movement throughout the body.

Much progress has been made in the quest for carriers ofionsand other small molecules, but the molecular mechanism oftransmembrane water movement has remained poorly un-derstood (1). Water-selective channels have been proposedto explain the large osmotic water permeability of erythro-cytes and certain epithelial cells (2), and the aquaporins, afamily of water-transporting proteins, were recently identi-fied in mammals and plants (3).

Discovery of the aquaporin CHIP [channel-forming inte-gral membrane protein of 28 kDa (CHIP28)] (4, 5) andisolation of its cDNA (6) permitted the demonstration of amolecular water channel. This was achieved by expression ofCHIP in Xenopus oocytes (7) and verified with syntheticproteoliposomes reconstituted with highly purified erythro-cyte CHIP (8). Similar to water channels in native mem-branes (1), CHIP-mediated water transport was reversiblyinhibited by HgCl2 reacting with Cys-189 in the CHIP mol-ecule (9). Immunoblot screening revealed abundant CHIP inkidney (4) where it comprises 4% of the total protein incortical brush-border membrane vesicles (10). Using affinity-purified antibodies specific for the N- and C-terminal do-mains, detailed immunohistochemistry and immunoelectron

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

microscopy localized CHIP in both apical and basolateralmembranes of proximal tubules and descending thin limbs ofHenle's loop, a distribution in exact concordance with theknown water permeability ofthe nephron (10). These findingshave been confirmed and further defined by others (11-13).Northern analyses revealed CHIPmRNA in several organs

(6, 14, 15). CHIP transcripts were demonstrated in diverseepithelia in fetal rats where three distinct patterns of expres-sion were identified (16). CHIP has been proposed to be themajor mechanism by which transmembrane water movementoccurs in mammals (3), and this study was undertaken todocument the tissue and cellular distributions of CHIPthroughout the body.

MATERIALS AND METHODSAntibodies to CHIP. Immunization of rabbits with highly

purified CHIP and affinity purification of antibodies wasdescribed (5, 10). Antibodies to human erythrocyte spectrinand band 3 were a gift from Vann Bennett (Duke UniversityMedical Center).

Tissue Preparation for Immunoblots. Four mature Sprague-Dawley rats were injected with 100 units of heparin and killedby ether inhalation, and the carcasses were perfused throughthe right and left ventricles with 500 ml of0.15M NaCl/7.5mMsodium phosphate/i mM sodium EDTA, pH 7.5. Selectedorgans were removed and homogenized in buffer containing0.25 M sucrose, 1 mM phenylmethylsulfonyl fluoride, diiso-propylfluorophosphate at 0.5 mg/ml, and leupeptin at 4 ,ug/mlat 0°C as described (4). Membranes were solubilized in SDS,electrophoresed into 12% polyacrylamide gels, and immuno-blotted with chemiluminescence autoradiography as described(10).

Immunolocalization of CHIP. Methods were derived fromour previous study (10). Albino Wistar rats were anesthetizedwith pentobarbital (5 mg/100 g) and fixed by vascular per-fusion through the left ventricle with 8% paraformaldehyde in0.15 M sodium cacodylate buffer. Tissue blocks were incu-bated 2 hr additionally before infiltration for 30 min in 2.3 Msucrose containing 2% paraformaldehyde and then weremounted on holders and rapidly frozen in liquid N2. Surgicalspecimens of human tissues immersion-fixed with 3% para-formaldehyde were provided by the Departments of Pathol-ogy, Aarhus University Hospital.

Light microscopy was performed on 0.85-,um cryosectionsor 2-,m paraffin sections after incubation with anti-CHIP(0.07-0.3 ,ug/ml), visualized with horseradish peroxidase-conjugated secondary antibody, and counterstained withtoluidine blue or Mayer's hematoxylin (10). Electron micros-copy was performed on 800-A ultrathin cryosections visual-ized with protein A-gold (10). Specificity of immunolabeling

Abbreviation: CHIP, channel-forming integral membrane protein of28 kDa (CHIP28).*To whom reprint requests should be addressed.

7275

Dow

nloa

ded

by g

uest

on

July

20,

202

0

Proc. Natl. Acad. Sci. USA 90 (1993)

was confirmed in all tissues with (i) nonimmune rabbit IgG,(ii) preadsorption of anti-CHIP with a >50-fold molar excessof purified CHIP, and (iii) incubation without primary anti-body.

RESULTSTissue Immunoblots. The presence of CHIP was estab-

lished in membranes isolated from several rat tissues byimmunoblotting with anti-CHIP, an affinity-purified IgG spe-cifilc for the C-terminal 3-kDa cytoplasmic domain of CHIP(5, 10). Strikingly abundant CHIP was noted in erythrocytesand kidney membranes, while smaller amounts of immuno-reactive CHIP were detected in other tissues except wholebrain (Fig. 1). The intensity of the immunoblot was strongerwith membranes from anterior chamber of eye, lung, andlactating mammary than with membranes from heart, choroidplexus, and liver (Fig. 1). Electrophoretic mobilities of gly-cosylated CHIP subunits from these tissues were distinctfrom erythrocyted glycosylated CHIP, and immunoblottingwith antibodies specific for spectrin or band 3 revealednegligible erythrocyte contamination.

Distribution of CHIP in Secretory and Resorptive Epithelia.Although anti-CHIP immunostaining was undetectable else-where in rat brain, choroid plexus exhibited striking immu-nostaining restricted to the apical microvilli of epithelial cellsin the villous-like processes; basolateral membranes andfenestrated capillaries failed to react (Fig. 2a). Specificity ofthe staining was demonstrated by preadsorption ofanti-CHIPwith purified CHIP (Fig. 2a-neg). Subcellular localization byimmunoelectron microscopy with anti-CHIP and proteinA-gold revealed immunoreactivity with the inner leaflets ofmicrovillar membranes but not with intracellular or basolat-eral membranes (Fig. 3a).

Coomassie

a) C>

1§|1

FglyCHIP-CHIP

Anti-CHIP

Anti-spectrin

Anti-band 3

FIG. 1. SDS/polyacrylamide gel electrophoresis and immuno-blotting of membranes isolated from various rat tissues. Samples ofmembrane proteins (10-50 Mg) were electrophoresed into 12% poly-acrylamide gel slabs and stained with Coomassie blue (Top) or

immunoblotted with anti-CHIP, anti-erythrocyte spectrin, or anti-erythrocyte band 3. Autoradiographic exposure time of red-cell andkidney lanes was reduced by 90% due to the higher signal intensity.glyCHIP, glycosylated CHIP.

*

,lb -a. ...

I.-

~m* *:

a - ra-neg

III44

itt

b-neg

..............t................

....t.._..

c c-neg

FIG. 2. Immunohistochemical localization of CHIP in choroidplexus and ocular epithelia. Rat tissue sections of 0.85 ,um (a and b')or 2 um (other panels) were incubated with anti-CHIP followed withperoxidase goat-anti-rabbit IgG and counterstain. (a) Choroid plexusepithelium from the fourth ventricle. Apical microvillar membranesexhibited intense immunoreaction (large arrows), whereas basolat-eral membranes (arrows) and capillaries (asterisk) were negative.(x775.) (a-neg) Preadsorbed control (anti-CHIP was preadsorbedwith pure CHIP). (X245.) (b-c) Anterior chamber of eye. (b) Epi-thelium covering ciliary body (CB), anterior and posterior epitheliumof iris (IR, small arrows), and anterior ciliary process (large arrow)all showed strong immunostaining; posterior ciliary processes werenegative. (X45.) (b') Higher magnification showing immunolabelingof apical and basolateral parts of iris epithelium (small arrows).(x200.) (b-neg) Preadsorbed control. (x27.) (b") Corneal endothe-lium (small arrows) was distinctly positive, as were stromal compo-nents surrounding tissue clefts (large arrows), but corneal epitheliumwas negative. (x90.) (c) Epithelium over anterior lens (arrows)exhibited immunoreaction with anti-CHIP. (x65.) (c-neg) Pread-sorbed control. (x36.)

Structures within the anterior chamber of rat eye reactedwith anti-CHIP. Iris epithelium and the anterior parts of theciliary body and processes were immunostained, but theposterior folds were not (Fig. 2b). Higher magnificationrevealed immunoreactive CHIP within the double-layerediris epithelium (Fig. 2b'), and immunoelectron microscopyrevealed CHIP in the apical and basolateral ciliary mem-branes (Fig. 3b). Corneal endothelium was labeled with

7276 Cell Biology: Nielsen et al.

Dow

nloa

ded

by g

uest

on

July

20,

202

0

Proc. Natl. Acad. Sci. USA 90 (1993) 7277

FV

\\- ...... . ..... . ^ *g I'.. ..

^ bx: .-. ......,.' . .. $w^tl ....... ,,:MV

lCs.. "p.z

,. '';. ..' ..'..."..',......

P1- r: f

* . ' .,, ..#glJ~

BLB.

.:>.:.:....

ll ! a § ~~~~~~~. .. .......... ' ',''J''''.......... Mi': i , .: .... : . ' k' tS,,t' ,'.k ' ~~~~......... ...IM|e>f7AE ;!SX; R

;5

OF*N s*tJ

sf :. :.: ...,'

,,,33p!'.-.'..,,^.,w,,,.,i...,.,...!.'''1!79 4i':s .. SF : , :, ....... :: . . ': : : .,, :: ; ........... '# # .:f.

..:.:..:: :..'.:.:..f. .'.:'. fs '. ':

BM

FIG. 3. Subcellular localization ofCHIP in rat tissues by immunoelectron microscopy with anti-CHIP and protein A-gold (10 nm). (a) Choroidplexus epithelium showed strong immunolabeling of the inner leaflet (arrows) of microvillar membranes (MV), whereas basolateral membranes(BL) and intracellular structures exhibited sparse immunolabeling. M, mitochondrion. (x29,250.) (b) Tangential section through basolateralmembranes from the inner layer of ciliary epithelium surrounding intercellular spaces (ICS) exhibited strong immunolabeling. (x35,100.) (Inset)Apical membranes oftwo cells (1 and 2) exhibited anti-CHIP immunolabeling. (x21,600.) (c) Apical and basolateral membranes (BM) of capillaryendothelium from skeletal muscle exhibited immunolabeling (arrows). (x43,200.) (d) Plasmalemmal membrane invaginations (caveolae) intangential section through capillary endothelium were also immunolabeled (arrows); the basement membrane is not well reproduced (BM).(x40,500.)

anti-CHIP, but corneal epithelium was unreactive (Fig. 2b").Also, both apical and basolateral membranes of lens epithe-lium contained immunoreactive CHIP, whereas lens fibercells failed to react (Fig. 2c). Specificity of these studies wasconfirmed with preadsorbed anti-CHIP (Fig. 2 a-, b-, andc-neg).

Sections from human liver ana neck of human gall bladderwere studied, since rats lack gall bladders. Hepatic sinusoidswere unreactive, but the continuous endothelium in peribil-iary capillary plexus (Fig. 4 a and a'), interlobular hepatic bileducts (Fig. 4a), and terminal bile ductules (Fig. 4a') were

strongly labeled with anti-CHIP. Anti-CHIP immunostainingwas weak over the apical membranes but strong over baso-lateral membranes of gall bladder (Fig. 4 b and b'). Pread-sorbed anti-CHIP failed to react with liver or gall bladder(Fig. 4 a- and b-neg). Distinct labeling of intestinal lymphaticvessels (lacteals) was observed, whereas epithelium andfenestrated capillaries in small intestine (Fig. 4 c and c') andcolon (Fig. 4d) were not labeled.

Localization of CHIP Within Capillary Endothelia. Thestrong reaction of immunoblots containing membranes ofseveral tissues is explained by the presence of CHIP reac-

tivity in capillary endothelium rather than with tissue paren-chyma. Lung contained abundant anti-CHIP immunoreac-tivity within capillaries surrounding bronchioles, alveoli, andvisceral pleura mesothelium, where the entire circumferenceof capillaries appeared to be immunolabeled. Although epi-thelial cells and smooth muscle in bronchiolar airways were

clearly not immunostained, reaction with connective tissueand smooth muscle elsewhere in the lung parenchyma cannotbe excluded (Fig. 5 a and a').The glandular epithelium of lactating mammary failed to

react with anti-CHIP, but capillaries in the surrounding con-nective tissue exhibited strong immunoreaction (Fig. 5b), adistribution shared by pancreas, lacrimal glands, and salivaryglands (data not shown). The capillary beds with continuousendothelium within skeletal, cardiac, and smooth muscle werealso strongly reactive with anti-CHIP, whereas the myofibrilswere unreactive (Fig. 5 c-e). Specificity of these capillaryimmunohistochemical studies was confirmed by the lack ofimmunostaining with preadsorbed anti-CHIP (Figs. 5 a- andd-neg). Immunoelectron microscopy confirmed CHIP immu-noreactivity over the apical and basolateral membranes ofskeletal muscle capillary endothelium as well as small plas-malemmal vesicular structures (Fig. 3 c and d).

DISCUSSIONThis study establishes the locations of CHIP in severalsecretory and resorptive epithelia and in continuous endo-thelia of capillaries. The results underscore the hypothesisthat CHIP plays a major role in transmembrane water flowthroughout the body and provides molecular insight into thisphenomenon which occurs in diverse physiologic and patho-logic processes.

CeU Biology: Nielsen et al.

...->.

-r 1:1.. . ....I -,M:N :- ..

I -.4. ."

IBMU7

.i.

.......

Dow

nloa

ded

by g

uest

on

July

20,

202

0

7278 Cell Biology: Nielsen et al.

.J'I! .s,9yt*' ) ;, ,

2~~~~~~~~~~~e ..:

ir \ s $bA ~f ts1 SL~~~~~'i,f8.S iN

"4~~~~~~~~~~~*#w i~~~~~~~~-

4..

0*4

T.&3,, ,//jb -

I

b-neg...

Proc. Natl. Acad. Sci. USA 90 (1993)

t @ *;i r:t^5.'**~~~~~~~~,kld anC@r eg

a

'4

~~~~r~~~~AC..;~~~9j'4s.. -jai - ,fa-tn S < .

>EtsS~~~ -9t- EPSLt . d*$ !t', v f i.:.:§ ;:.E S

¢| s..SE.Y,; .;;,,....................... fiS4s :,,. ....s,Ss.LW

FIG. 4. Immunohistochemical localization of CHIP in 2-,um paraffin sections of human liver and gall bladder and 0.85-/am frozen sectionsoffixed rat intestine. (a-a) Human liver interlobular bile ducts (large arrows, a) and bile ductules (large arrows, a') and nonfenestrated peribiliarycapillaries (small arrows) were all distinctly immunolabeled; no labeling of sinusoids was observed in periportal areas (arrowheads, a' and a").(x200, x260, and x350, respectively.) (a-neg) Preadsorbed control. (x90.) (b) Neck of human gall bladder showed weak immunoreaction ofthe apical membrane and strong immunoreaction ofthe lateral membrane (arrows). (x90.) (b') Higher magnification. (x200.) (b-neg) Preadsorbedcontrol. (x23.) (c and c') Central lymphatics (lacteals, L) within duodenum (c) and ileum (c') exhibit labeling, but fenestrated capillaries(arrowheads) and the epithelium (EP) do not. (x890 and x300.) (d) Rat colon mucosal epithelium (EP) and capillaries (arrowheads) exhibitedno immunoreactivity, whereas the central structure (probably a lymphatic, arrow) reacted strongly with anti-CHIP. (x390.)

CHIP was identified at only a single site in brain, the apicalmicrovillar membrane of the choroid plexus epithelium.Na+,K+-ATPase is also exclusively located there (17), pro-

viding release of Na+ into the subarachnoid space whichdrives the passive movement ofwater into cerebrospinal fluid(18). CHIP was identified at both apical and basolateral

* 7 IJ; ; k

- 1k'0. isC I,

A 7 1 j

'0? /

'0. 0 * '0 'A

C,1' i

* F,/.

if e

r.

A

'a

_.DN

: d

_n~~~~~~~~~~~~~~~~~Ql0ji

.d-neg..

A '** -'..- t .....;-v

.r2:,, :'AS i;L ^

aLneg r ,_ $;t¢Vb#N * >S .I

I

1.:- I

I., ..

*1 . .

* V....6:

d.I )

FIG. 5. Immunohistochemical localization of CHIP within capillary endothelia. Strong immunoreactions with anti-CHIP were observed incapillaries (arrows) in 2-,um (a and c) or 0.85-pm (b, d, and e) sections of rat tissues. (a) Cross section of bronchiole and capillary bed. (x80.)(a') Respiratory section of lung. (x 180.) (a-neg) Preadsorbed control. (x57.) (b) Lactating mammary gland with capillaries surrounding alveolarglandular epithelium (A). (x540.) (c) Skeletal muscle. (x80.) (d) Skeletal muscle at higher magnification. (x270.) (d-neg) Preadsorbed control.(xlO0.) (e) Myocardium. (x270.)

A.

b j

.4. ..

Y*r

Dow

nloa

ded

by g

uest

on

July

20,

202

0

Proc. Natl. Acad. Sci. USA 90 (1993) 7279

membranes of the ciliary epithelium in the eye, where aque-ous humor is secreted (19). CHIP is present in the anteriorepithelium of lens and in the corneal endothelium, wherefunctional water channels apparently maintain desiccationand transparency of these ocular structures (20). CHIP waslocalized in both apical and basolateral membranes of intra-hepatic bile ductules and ducts, where bile is secreted, andgall bladder epithelium, where bile is concentrated (21).Abundant CHIP exists in apical and basolateral membranesof continuous capillaries of several tissues but not in theparenchyma of muscle or various exocrine glands. Thesecapillaries are known to be water-permeable, balancing theopposing forces of hydrostatic pressure, which drives waterinto the interstitium, and oncotic pressure, which drawswater back into the vascular space (22).The molecular basis of water transport in each of these

phenomena is unsettled, but suggested explanations includetransjunctional pores and transmembrane diffusional mech-anisms. We propose that CHIP functions in the aforemen-tioned epithelia by conferring high osmotic water permeabil-ity responsible for (i) secretion of water into cerebrospinalfluid and aqueous humor, (ii) removal of water from corneaand lens, (iii) secretion of water into bile by intrahepaticbiliary ductules and resorption of water from the gall bladderas bile is concentrated, and (iv) water release and reuptake bycontinuous capillaries, which maintain interstitial fluid vol-ume or airway humidity. These studies may also illuminateknown pathological processes including the development ofeffusions or edema due to changes in hydrostatic or oncoticpressures. The abundance of CHIP in the rich plexus sur-rounding alveolar spaces may explain the rapid water uptakeand hemolysis of freshwater drownings.CHIP is not located at all points where transmembrane

water movements occur, and therefore other mechanismsmust exist. Fenestrated capillaries probably contain struc-tural pores, and CHIP was not observed in fenestratedcapillaries in kidney (10), choroid plexus, lamina propria ofintestine, or elsewhere. To avoid hemolysis, gradual waterabsorption through gastrointestinal mucosal epithelium mustoccur, and physiologic measurements have demonstratedlow water permeability of intestinal epithelium (23), whichwas shown here to lack CHIP. Intestinal water absorptionmay therefore occur by diffusional or paraceilular mecha-nisms with rapid dilution into vascular spaces driven by theincreased oncotic pressures within the lumens of the intes-tinal lacteals and capillaries. CHIP was not detected in therenal collecting ducts (10), where the vasopressin-regulatedwater channel exists (24), or in basolateral membranes ofchoroid plexus epithelium, in the posterior ciliary processes,or in membranes of several secretory glands, including mam-mary, salivary, and lacrimal glands. Although this may reflectinaccessibility of antigen, it is more likely that other waterchannels exist in some of these locations.

We thank the manuscript reviewers for important observations and

detailed suggestions for presentation of data. H. Sidelman and I.Kristoffersen provided excellent technical assistance. Support forthis study was provided by the Danish Medical Research Council, theDanish Foundation for Advancement of Medical Science, the Bio-membrane Research Center and Research Foundation of the Uni-versity of Aarhus, the Novo Foundation, and National Institutes ofHealth Grants HL33991 and HL48268.

1. Finkelstein, A. (1987) Water Movement Through Lipid Bilay-ers, Pores, and Plasma Membranes, Theory and Reality (Wi-ley, New York).

2. Solomon, A. K., Chasan, B., Dix, J. A., Lukacovic, M. F.,Toon, M. R. & Verkman, A. S. (1983) Ann. N. Y. Acad. Sci.414, 97-124.

3. Agre, P., Sasaki S. & Chrispeels, M. J. (1993) Am. J. Physiol.,in press.

4. Denker, B. M., Smith, B. L., Kuhajda, F. P. & Agre, P. (1988)J. Biol. Chem. 263, 15634-15642.

5. Smith, B. L. & Agre, P. (1991) J. Biol. Chem. 266, 6407-6415.6. Preston, G. M. & Agre, P. (1991) Proc. Natl. Acad. Sci. USA

88, 11110-11114.7. Preston, G. M., Carroll, T. P., Guggino, W. B. & Agre, P.

(1992) Science 256, 385-387.8. Zeidel, M. L., Ambudkar, S. V., Smith, B. L. & Agre, P.

(1992) Biochemistry 31, 7436-7440.9. Preston, G. M., Jung, J. S., Guggino, W. B. & Agre, P. (1993)

J. Biol. Chem. 268, 17-20.10. Nielsen, S., Smith, B. L., Christensen, E. I., Knepper, M. A.

& Agre, P. (1993) J. Cell Biol. 120, 371-383.11. van Hoek, A. N. & Verkman, A. S. (1992) J. Biol. Chem. 267,

18267-18269.12. Zhang, R., Skach, W., Hasegawa, H., van Hoek, A. N. &

Verkman, A. S. (1993) J. Cell Biol. 120, 359-369.13. Sabolic, I., Valenti, G., Verbavatz, J.-M., van Hoek, A. N.,

Verkman, A. S., Ausiello, D. A. & Brown, D. (1992) Am. J.Phys. 263, C1225-C1233.

14. Lanahan, A., Williams, J. B., Sanders, L. K. & Nathans,D. M. (1992) Mol. Cell. Biol. 12, 3919-3929.

15. Deen, P. M. T., Dempster, J. A., Wieringa, B. & Van Os,C. H. (1992)Biochem. Biophys. Res. Commun. 188, 1267-1273.

16. Bondy, C., Chin, E., Smith, B. L., Preston, G. M. & Agre, P.(1993) Proc. Natl. Acad. Sci. USA 90, 4500-4504.

17. Ernst, S. A., Palacios, J. R., II, & Siegel, G. J. (1986) J.Histochem. Cytochem. 34, 189-195.

18. Davson, H., Welch, K. & Segal, M. B. (1987) The Physiologyand Pathophysiology of the Cerebrospinal Fluid (ChurchillLivingstone, Edinburgh), pp. 15-34.

19. Bill, A. (1984) in Handbook ofPhysiology, eds. Renkin, E. M.& Michel, C. C. (Am. Physiological Soc., Bethesda, MD), pp.1001-1034.

20. Narula, P., Xu, M., Kuang, K. Y., Akiyama, P. & Fischbarg,J. (1992) Am. J. Physiol. 262, C98-C103.

21. Anwer, M. S. (1991) Prog. Pharmacol. Clin. Pharmacol. 8,3-23.

22. Michell, C. C. (1992) in The Kidney: Physiology and Patho-physiology, eds. Seldin, D. W. & Giebisch, G. (Raven, NewYork), 2nd. Ed., pp. 61-91.

23. van Heeswjk, M. P. E. & van Os, C. H. (1986)J. Membr. Biol.92, 183-193.

24. Fushimi, K., Uchida, S., Hara, Y., Hirata, Y., Marumo, F. &Sasaki, S. (1993) Nature (London) 361, 549-552.

Cell Biology: Nielsen et al.

Dow

nloa

ded

by g

uest

on

July

20,

202

0